text stringlengths 0 2.35k |
|---|
**Break:** \[21:50\] |
**Mat Ryer:** I love that advice of "Pay attention to the value you're gonna get from the effort that you put in." I like monorepos - I'd just like to put that out there; I love monorepos. And the reason I like them is because you can have a pull request that has a unit test, some backend code, maybe some API changes, ... |
**Nayana Shetty:** \[23:59\] I would love to say yes, but I've not seen a team do it really well. I can see the challenges -- like, when you have this monorepo and everyone is contributing to the same central repository, there is a challenge that the parameters that you would think about for your product and your monit... |
But on a single individual product team's perspective - yeah, I don't know how much value it would add, so it depends on that, I guess. I'm not sure. Have you seen it work in your teams, or something? |
**Mat Ryer:** Well, we have at least the conversation... When there's a PR for like a big feature, we will chat about it, and say "What do we need from this? What's going on here that later we're gonna need?" And it is that thing about "Be kind to your future selves." But I don't know that we've got that right yet, or ... |
**Nayana Shetty:** I agree, yeah. |
**Matt Toback:** All this to me starts to distill down into -- it's some amount of like if you are doing the centralized monitoring, or there's a level of that, and then you have to communicate this down to these teams and you have to get them to buy in... Where do you do that? Or even how would you suggest someone els... |
**Nayana Shetty:** Central teams pushing things is -- like, irrespective of it being monitoring or anything in general is really hard. It should always be driven by -- like, what I've seen work really well is the ones that are driven by value-add to the individual team itself. |
As an example, when we were building this Amazon Linux, like a baseplate image that everyone could apply, and they can run their own EC2 instances. When they had this, what we said we will do as part of it is we said "You're going to get monitoring --" I think we were pushing logs to Splunk in that case. So you would g... |
**Matt Toback:** Right. Make it so easy that they would rather adopt it, rather than trying to do it themselves. |
**Nayana Shetty:** Yeah, exactly. Another example that came to mind was we had this central repository for -- like a CRM system which we had to enter all of our system information in. Basically, it was like a -- because we had so many microservices, we had a central system where you could go in and query for any partic... |
And what we did when we built this - we said "If you put the right information in this, then you would automatically have a dashboard that would show up only your team's monitoring in it." That was an incentive for teams to be like "Oh, if I did this, then I get my own dashboard. Let me do that." |
\[27:47\] So I think it's that showing intrinsic value beyond just what you want them to achieve out of it. That's how I've seen it work really well in teams. So yeah, you need to have some sort of carrots to actually get people to move towards your solutions and stuff. |
**Matt Toback:** Yeah, yeah. |
**Mat Ryer:** Yeah, that's great. I think that applies to... Everything! \[laughter\] Yeah, if you make it easy and it's sort of a no-brainer... And one example of that is where we can -- like, if we've got APIs, we can just instrument on the endpoints very easily, in a simple way, usually with some middleware or somet... |
**Nayana Shetty:** Yeah. And I think it's the -- I would really want to see central teams be more mindful about this... Because as a central team, you're building these amazing tools, and at the end of it you kind of think "Oh, if I put a documentation together and self-service, everyone's gonna come and use it." But t... |
And also, it's that education piece of "You care about your product. We will help you care about your product." That's something to think about, I guess. |
**Matt Toback:** Yeah. |
**Mat Ryer:** So what are some common mistakes that we make when we're trying to do this? With the best intentions in the world, we want to do this properly, but are there any things you see that people misunderstand, or common mistakes, common gotchas that you've seen? |
**Nayana Shetty:** I think knowing how much is enough is one of the things I've often seen, where there are teams who just put the basic thing available because it's there in a checklist somewhere, and then they move on... Which is probably not the best for your product. So it's being aware of the value of your product... |
The other thing that I have seen and I've struggled a lot with is, like I mentioned, about this USE method and RED method to actually build your dashboards. It's very hard to get your network-related monitoring right, and the saturation for networks... Like, how do you do that? And get the wrong set of -- like, I've se... |
So I think it's just being okay to experiment and continuously tinker your monitoring and alerting as you go along is probably something that teams should be conscious that -- like, it's not that you build it once and then it's there forever, but there is a continuous evolution that happens with your monitoring... Like... |
**Mat Ryer:** Yeah, yeah, as -- |
**Matt Toback:** Mat, can I answer, too? |
**Mat Ryer:** Let me just check... No. \[laughter\] |
**Matt Toback:** Oh, come on...! |
**Mat Ryer:** Brutal, but... \[laughs\] Yeah, please, I'd love to hear what you think, of course. |
**Matt Toback:** Okay. So Nayana, you were thinking about the value derived, and focusing on that for the customers... I do think that's a common gotcha, where you build all these tools and you're like "We did it! We did it! It's all there. All you have to do is this." |
\[31:46\] I think the common gotcha is forgetting that you need to deliver something that someone can just adopt easily, like you said. It is a version of -- I was thinking like car parts, or legos, I guess... Dropping off like a collection of car parts and being like "There you go!" And you're like "I wanna drive. I g... |
But I do think there's some version of that too, stopping short of actually delivering the value to the person consuming it, as opposed to just dropping a collection of pieces that can work, but they have to do the last mile. |
**Mat Ryer:** Well, in a way, what helps that definitely is gonna be this "You build it and you run it." We're not throwing this thing over the wall for someone else to operate... Which I know that actually lots of people do still do that. And there's a disconnect. When you are yourselves kind of running it, you're the... |
**Nayana Shetty:** Yeah. And also, I think one of the comments I've heard a few people say about this - build your code in such a way that you can debug it at three in the morning. It doesn't mean that you have to do it every day, but if it breaks at a time that you're not fully in focus, you still can get to it easily... |
**Mat Ryer:** That's such a great point, I think... And that leads me to our next question, which is around drills. Should we be doing drills at 3 AM, and living that experience to see what it's like? |
**Nayana Shetty:** Three o'clock is probably taking the Mickey out of people if you were doing drills... \[laughter\] |
**Matt Toback:** Do people do drills? I guess they do, but it's probably not common, is it? |
**Nayana Shetty:** I have seen it done, and I think it's a very artificial environment where the drills happen. One of the things we did when I was at FT was we had these incident drills. Basically, you emulate an incident and then you go about with the team "How do you go about actually figuring out where the problem ... |
**Mat Ryer:** It's because you didn't at 3 AM. \[laughter\] |
**Nayana Shetty:** Yeah, maybe that. |
**Matt Toback:** I think there's like a touch of maturity in actually embracing drills, whether or not it's artificial. It's that idea "Oh, this is artificial. This is dumb. We don't wanna do this. It's not gonna be like this in real life." And then you think about any kind of -- I don't know, any team environment or a... |
**Mat Ryer:** Yeah, it's not the same... It's not the same, because you know it's a drill. Unless you're doing something where you literally break something and it's not really broken... Or maybe it is, and you're doing something kind of "That seems a bit extreme..." It is gonna feel different, but that still doesn't m... |
**Matt Toback:** \[36:03\] You press the horn to go... \[laughter\] |
**Mat Ryer:** Yeah, horn to go, and then you leave that on, so that everyone knows you're there. I do, because they need to get out of the way... \[laughs\] |
**Nayana Shetty:** I think there's also value in -- the other way of looking at drills is like shadowing, and when there's an actual incident not having just one or two people involved in it. Yes, it might be the most critical thing, but having more people just listening and see what's happening, and just be there some... |
**Matt Toback:** I even think yesterday - and I realize why I'm all fired up about this... I parsed through it. Yesterday I visited the sales team, and they were doing these workshops, and they were doing Radical Candor, which is all about feedback, and giving feedback, and getting feedback, and being able to do it wel... |
**Mat Ryer:** I think that was a rubbish point. |
**Matt Toback:** We can cut that. Cut, cut, cut. \[laughter\] |
**Nayana Shetty:** I think it's also a good exercise to do, just to test your documentation, if your documentation is up to scratch. When you've written something, you've written with good intent, but when someone's actually following it, does it make sense? It's something that the drills can actually capture. There's ... |
**Break:** \[38:04\] |
**Mat Ryer:** Nayana, you mentioned earlier this idea that if you do too much, you can overdo it and end up with basically alert fatigue... Just alerts going off. What do we mean really by alert fatigue? |
**Nayana Shetty:** I'm going to give an example so people can relate to it. I was in one of the teams where we used to get close to 1,500 alerts on a weekly basis... We had around 80-odd microservices. So it wasn't like just one microservice, or anything. But then my team was three people looking at this, and it's at t... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.